80 likes | 242 Views
3D Face Recognition Using Range Images. Literature Survey Joonsoo Lee 3/10/05. intensity image 3D mesh range image. Introduction. Face Recognition Develop an automatic system which can recognize the human face as humans do Image data 2D: intensity image 3D: mesh, range image
E N D
3D Face Recognition Using Range Images Literature Survey Joonsoo Lee 3/10/05
intensity image • 3D mesh • range image Introduction • Face Recognition • Develop an automatic system which can recognize the human face as humans do • Image data • 2D: intensity image • 3D: mesh, range image (a) (b) (c)
Image Acquisition Classification Feature Extraction Pre-processing Capture face images & Generate range images Design a classifier, train it with dataset, and test its validity Extract the features from normalized face images Normalize images into the same position Background • Range Image • Image with depth information • Invariant to the change of illumination & color • Simple representation of 3D information • Procedure
Geometrical Approach • Principal Curvature [Gordon (1991)] Method • Calculate principal curvatures on the surface • Generate face descriptors from curvarture map Remark Outline of the use of curvature information in the process of face recognition Can deal with faces different in size Advantage Disadvantage Need some extension to cope with changes in facial expression
Geometrical Approach • Spherical Correlation [Tanaka & Ikeda (1998)] Method • Construct Extended Gaussian Image (EGI) • Compute Fisher’s spherical correlation on EGI’s Remark First work to investigate and evaluate free-formed curved surface recognition Advantage Simple, efficient, and robust to distractions such as glasses and facial hair Disadvantage Not tested on faces in different sizes and facial expressions
Advantage Disadvantage Large dimension reduction Bad performance with large database Statistical Approach • Eigenface [Achermann et al. (1997)] Method • Consider face images as vectors • Apply principal component analysis (PCA) Remark • Optimal in the least mean square error sense • Prevalent method in 2D face recognition [Turk & Pentland (1991)]
Disadvantage Lots of computation due to optimization problem Statistical Approach • Optimal Linear Component [Liu et al. (2004)] Method • Consider face images as vectors • Find optimal linear subspaces for recognition Optimal in the sense that the ratio of the between-class distance and within-class distance is maximized Remark Better performance than standard projections, such as PCA, ICA, or FDA Advantage